referee
Pass
Audited by Gen Agent Trust Hub on Oct 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external content such as code diffs, pull requests, and written acceptance criteria. This creates a surface where malicious instructions embedded within the reviewed code could attempt to manipulate the agent's reasoning or findings.
- Ingestion points: Processes diffs, criteria, and repository state as primary inputs (SKILL.md).
- Boundary markers: The instructions do not mandate the use of specific delimiters or explicit 'ignore embedded instructions' warnings for the external data being analyzed.
- Capability inventory: Employs subagents and repository search tools to verify context and code signatures (SKILL.md).
- Sanitization: No mechanisms for sanitizing or escaping the ingested content are specified in the workflow.
Audit Metadata